1. automotive and vehicles
2. cars
3. crossover

WSEAS TRANSACTIONS on COMPUTERS
Jorge Magalhães-Mendes
A Comparative Study of Crossover Operators for Genetic Algorithms to
Solve the Job Shop Scheduling Problem
JORGE MAGALHÃES-MENDES
Department of Civil Engineering
CIDEM
School of Engineering – Polytechnic of Porto
Rua Dr. António Bernardino de Almeida, 431 – 4200-072 Porto
PORTUGAL
jjm@isep.ipp.pt
Abstract: - Genetic algorithms (GA) are wide class of global optimization methods. Many genetic algorithms
have been applied to solve combinatorial optimization problems. One of the problems in using genetic
algorithms is the choice of crossover operator. The aim of this paper is to show the influence of genetic
crossover operators on the performance of a genetic algorithm. The GA is applied to the job shop scheduling
problem (JSSP). To achieve this aim an experimental study of a set of crossover operators is presented. The
experimental study is based on a decision support system (DSS). To compare the abilities of different crossover
operators, the DSS was designed giving all the operators the same opportunities. The genetic crossover operators
are tested on a set of standard instances taken from the literature. The makespan is the measure used to evaluate
the genetic crossover operators. The main conclusion is that there is a crossover operator having the best average
performance on a specific set of solved instances.
Key-Words: - Scheduling, Genetic Algorithms, Crossover Operators, Optimization, Operations Research, JSSP.
1515 are still considered to be beyond the reach of
today's exact methods.
Many approximate methods have been developed
in the last two decades to solve the JSSP, such as:
1. Simulated annealing (SA) - Lourenço [12];
2. Tabu search (TS) (Nowicki and Smutnicki
[9, 31], Pezzela and Merelli [7], Zhang et al.
[11, 35]),
3. Evolutionary algorithm (EA) techniques like
genetic algorithms (GA) - Aarts et al. [15],
Croce et al. [17], Dorndorf et al. [18], Wang
and Zheng [19], Essafi et al. [3], Gonçalves
et al. [4], Hasan et al. [27], Mendes [30],
Qing-dao-er-ji and Wang [38], Choi and Park
[39], Chiu et al. [40, 41]);
4. EA related techniques like particle swarm
optimization (PSO) - Sha and Hsu [6];
5. Greedy
randomized adaptive
search
procedure (GRASP) - Aiex et al. [16] and
Binato et al. [20];
6. Others metaheuristics - Rego and Duarte [8]
proposed a filter-and-fan approach based on
the shifting bottleneck procedure (SBP) and
Pardalos et al. [32].
1 Introduction and Background
Scheduling is one of the most critical issues in the
planning and managing of manufacturing process.
The difficulty of finding the optimal schedule
depends on the shop environment, the process
constrainsts and the performance indicator. One of
the most difficult problems in this area is the job shop
scheduling problem (JSSP), Qing-dao-er-ji and Wang
[38].
The JSSP may be described as follows: given n
jobs, each job must be processed on m machines.
Each operation uses one of the m machines for a
fixed duration. Each machine can process at most one
operation at a time and once an operation initiates
processing on a given machine it must complete
processing on that machine without interruption. A
schedule is a complete set of operations (n x m), to be
processed on different machines, in a given order.
The problem is to find a schedule of minimal time to
complete all jobs.
The JSSP is considered as a particularly hard
combinatorial optimization problem, Lawler et al.
[10].
Exact methods (Giffler and Thompson [21],
Carlier and Pinson [22, 23], Brucker et al. [24],
Williamson et al. [25]) have been successful in
solving small instances. Problems of dimension
E-ISSN: 2224-2872
Evolutionary algorithms often perform well
approximating solutions to all types of problems. The
most popular type of EA is the GA.
164
Issue 4, Volume 12, April 2013
WSEAS TRANSACTIONS on COMPUTERS
Jorge Magalhães-Mendes
{o0, o1, …, oji,… onm, onm+1}. Set O contains all the
operations of each job. Each job has m operations.
Each machine can process at most one operation at
time.
The JSSP is to find a schedule which minimizes
the makespan (Cmax), that is, the finish time of the last
operation completed in the schedule, taking into
account the precedence constraints.
Let O = {o0, o1, …, oji,… onm, onm+1}denote the
set of all operations (n x m) to be scheduled and M =
{1,..., m} the set of machines. The operations o0 and
onm+1 are dummy, have no duration and represent the
initial and final operations. The operations are
interrelated by two kinds of constraints:
 First, the precedence constraints, which force
each operation oji to be scheduled after all
predecessor operations are completed Pji;
 Second, operation oji can only be scheduled if
the machine it requires is idle. Further, let dji
denote the (fixed) duration (processing time)
of operation oji.
The GA seeks the solution of a problem in the
form of strings of numbers (traditionally binary,
although the best representations are usually those
that reflect something about the problem being
solved), by applying operators such as crossover and
mutation.
The performance of a GA is dependent on the
genetic operators in general and on the type of
crossover operator, in particular. During the evolution
process by a GA, if the selected chromosomes are
identical, some of the crossover operators have failed
to create offspring that are different from their
parents. Effective crossover in a GA is achieved
through establishing the optimum relationship
between the crossover and the search problem itself
[28].
Some literature focuses on the genetic operators,
e.g., Sywerda [33] and De Jong and Spears [34]
compared the various crossover operators,
particularly the numbers of crossover points. A
relevant result is that sometimes the final output
would be better if the number of crossover points was
increased.
Besides empirical analysis, substantial efforts
have been invested in comparing, from theoretical
perspectives, between mutation and crossover as well
as between the various crossover operators [34].
However, these theories are not general enough to
allow for predicting when to apply or what type of
crossover operator to employ [28].
Given the current state of knowledge is justified,
therefore, additional research work that allows to
obtain new results.
This paper describes an experimental study of a
set of genetic crossover operators. These crossover
operators are applied on a hybrid genetic algorithm.
This hybrid genetic algorithm works with a local
search using the Monte Carlo method [30]. The aim
of this study was to validate empirically the most
appropriate crossover operator for solving the job
shop scheduling problem.
The remainder of the paper is organized as
follows. In section 2 is defined the problem. In
section 3 is referred the decision support system. In
section 4 we refer the advantages of using genetic
algorithms in combinatorial nature problems.
In section 5 is described the genetic algorithm. In
section 6 are presented and discussed the results of
the experimental study. Finally, conclusion and
remarks for further works are given in Section 7.
Let Fji represent the finish time of operation oji. A
schedule can be represented by a vector of finish
times (F11, , Fji, ... , Fnm+1).
Let A(t) be the set of operations being processed
at time t, and let rji = 1 if operation oji requires
machine m to be processed and rji = 0 otherwise.
The conceptual model of the JSSP can be described
the following way:
Min Fnm 1
subject to:
Fkl  Fji  d ji

ji  A t 
rji  1
Fji  0
j  1,..., n ; i  1,..., m ; kl  Pji
(2)
iM ; t  0
(3)
j  1,..., n ;
i  1,..., m
(4)
The objective function (1) minimizes the finish
time of operation onm+1 (the last operation), and
therefore minimizes the makespan. Constraints (2)
impose the precedence relations between operations
and constraints (3) state that one machine can only
process one operation at a time. Finally (4) forces the
finish times to be non-negative.
3 A decision support system (DSS)
A DSS combines a genetic algorithm, a schedule
generator scheme (SGS) and a local search
procedure, consisting in the following steps, see
Figure 1:
 Step1: Combines a genetic algorithm
with a schedule generation scheme
2 Problem Definition
There are a set of jobs J = {1,…, n}, a set of
machines M = {1,..., m}, and a set of operations O =
E-ISSN: 2224-2872
(1)
165
Issue 4, Volume 12, April 2013
WSEAS TRANSACTIONS on COMPUTERS

Jorge Magalhães-Mendes
(SGS).
This
SGS
generates
parameterized active schedules. This step
allows to obtain a schedule for each
chromosome;
Step2: This step makes use of a local
search procedure that attempts to
improve the solution obtained previously.
genetic programming, evolutionary strategies,
neuroevolution and differential evolution.
The genetic algorithms have been applied
successfully in several areas, such as bioinformatics,
computational science, engineering, economics,
chemistry, manufacturing, mathematics and physics.
A real-coded GA is adopted in this article.
Compared with the binary-code GA, the real-coded
GA has several distinct advantages, which can be
summarized as follows, Y.-Z. Luo et al. [36]:
 It is more convenient for the real-coded GA
to denote large scale numbers and search in
large scope, and thus the computation
complexity is amended and the computation
efficiency is improved;
 The solution precision of the real-coded GA
is much higher than that of the binary-coded
GA;
 As the design variables are coded by floating
numbers in classical optimization algorithms,
the real-coded GA is more convenient for
combination with classical optimization
algorithms.
GA – decoding
SGS – Schedule generation scheme
Step 1
Schedule
Local search improvement
5 Genetic Algorithms
Step 2
Genetic algorithms (GA) are search algorithms based
on the mechanics of natural selection and natural
genetics. They combine survival of the fittest among
string structures with a structured yet randomized
information exchange to form a search algorithm with
some of the innovative flair of human search [1].
The general schema of GA may be illustrated as
follows (Fig. 2).
Improved schedule
Fig.1 – Architecture of the approach.
The DSS allows the user choose the crossover
operator:
 GA-MC-SPc: Single-Point crossover or
simple crossover;
 GA-MC-TPc: Two-Point crossover;
 GA-MC-Uc: Uniform crossover or discrete
crossover;
 GA-MC-Fc: Flat crossover.
procedure GENETIC-ALGORITHM
Generate initial population P0;
Evaluate population P0;
Initialize generation counter g 0;
While stopping criteria not satisfied repeat
Select some elements from Pg to copy into Pg+1;
Crossover some elements of Pg and put into Pg+1;
Mutate some elements of Pg and put into Pg+1;
Evaluate some elements of Pg and put into Pg+1;
Increment generation counter: g  g+1;
End while
The DSS was developed using the Visual Basic
for Applications (VBA) from Microsoft and the Gantt
Time Package V3.21 [30].
End GENETIC-ALGORITHM;
Fig.2 - Pseudo-code of a genetic algorithm.
4 Real-coded evolutionary algorithms
First of all, an initial population of potential
solutions (individuals) is generated randomly. A
selection procedure based on a fitness function
enables to choose the individuals candidate for
reproduction. The reproduction consists in
recombining two individuals by the crossover
The evolutionary algorithms are an interdisciplinary
research area comprising several paradigms inspired
by Darwinian principle of evolution.
The current stage of research considers, among
others, the following paradigms: genetic algorithms,
E-ISSN: 2224-2872
166
Issue 4, Volume 12, April 2013
WSEAS TRANSACTIONS on COMPUTERS
Jorge Magalhães-Mendes
operator, possibly followed by a mutation of the
offspring. Therefore, from the initial population a
new generation is obtained. From this new
generation, a second new generation is produced by
the same process and so on. The stop criterion is
normally based on the number of generations.

5.1 Decoding
The genetic algorithm uses a random key alphabet
which is comprised of real random numbers between
0 and 1.
A chromosome represents a solution to the
problem and is encoded as a vector of random keys
(random numbers).
Each solution chromosome is made of 2n genes
where n is the number of operations:
Usually, there are two means of modifying
genetic material: a recombination operation that
could be understood as some kind of crossover and
mutation.
In this study we compared the following
crossover operators:
 Single-point Crossover or simple crossover;
 Two-point Crossover;
 Uniform Crossover or discrete crossover;
 Flat Crossover.
Chromosome = (genel , .., genen , gene n+1 , ... , gene 2n )
The first half (genel, .., genen) corresponds to the
potential solution to the optimization problem, that is, one
gene for each activity. The second half (genen+1, .., gene2n)
defines the vector where each gene has an associated delay
time.
The reasons by which these crossover operators
were chosen are:
 Single-point crossover: is a simple and oftenused method for genetic algorithms;
 N-point crossover. Instead of only one, N
breaking points are chosen randomly.
Every
(1)
second section is swapped. Among this class,
two-point crossover is particularly important;
 Usually flat or discrete crossovers are applied
in real-coded genetic algorithm (Shaparov
[37]).
The priority decoding expression used the
following expression:
PRIORITY j  gene j
Delay j  genen j 1.5  mdur ,
effectiveness of crossover is depended on
coding;
Mutation (Asexual): In real evolution, the
genetic material can by changed randomly by
erroneous reproduction or other deformations
of genes, e.g. by gamma radiation. In genetic
algorithms, mutation realized as a random
deformation of alleles with a certain
probability, Shaparov [37].
j  1,..., n.
where mdur is the maximum duration of all
operations. The factor 1.5 was chosen as a result of a
careful fine-tuning experimental phase.
5.3.1 Single-point crossover
After reproduction, crossover may proceed in two
steps. First, members of the newly reproduced
chromosomes in the mating pool are mated at
random. Second, each pair of chromosomes
undergoes crossover as follows: an integer position k
along the chromosome is selected uniformly at
random between 1 and the chromosome length l. Two
new chromosomes are created swapping all the genes
between k+1 and l [1], see Fig. 3.
5.2 Initial population
The initial populations are generated randomly. The
quality of this population is poor and one way to
improve it is to incorporate some chromosomes
generated by priority rules.
In this paper are selected the priority rules GRPW
(greatest rank positional weight) and SPT (shortest
processing time) to improve some chromosomes of
the initial population.
5.3 Crossover operators
The genetic algorithms typically use the following
types of operators:
 Selection: Operator for selecting individuals
for reproduction according to their fitness;
 Crossover
(Sexual
Recombination):
Operator of merging the genetic information
of two individuals. In many respects the
Fig.3 – Single-point crossover operator example.
E-ISSN: 2224-2872
167
Issue 4, Volume 12, April 2013
WSEAS TRANSACTIONS on COMPUTERS
Jorge Magalhães-Mendes
5.3.2 Two-point crossover
Two-point crossover is very similar to single-point
crossover except that two cut-points are randomly
generated instead of one, see Fig. 4.
Second offspring is computed analogously, see Fig.6.
Fig.4 – Two-point crossover operator example.
Fig.6 – Flat crossover operator example.
5.3.3 Uniform crossover
In uniform crossover, a value of the first parent’s
gene is assigned to the first offspring and the value of
the second parent’s gene is to the second offspring
with a probability value pc.
With probability pc the value of the first parent’s
gene is assigned to the second offspring and the value
of the second parent’s gene is assigned to the first
offspring, see Fig. 5.
5.4 Random mutation
The mutation operator preserves diversification in the
search. The mutation operator chosen was the
random mutation. This operator is applied to each
offspring in the population with a predetermined
probability. For a randomly chosen gene i of an
individual (genel , .., genen , gene n+1 , ... , gene 2n ), the
allele genei is replaced by a randomly chosen value
from a interval ]0, 1[.
We assume that the probability of the mutation in
this work is 0.1%. With 1000 genes positions we
should expect 1000 x 0.001 = 1 genes to undergo
mutation for this probability value.
5.5 Configuration of the experiments
All results reported for GA-MC were obtained with
the same parameters. Though there is no
straightforward way to configure the parameters of a
genetic algorithm, we obtained good results with
values:
 Population size: 2 × number of operations
for each problem;
 Initial population: 1% chromosomes
calculated by priority rules GRPW and SPT;
 Top (best): 1% from the previous population
chromosomes are copied to the next
generation;
 Mutation rate: 0.1%;
 Termination criterion: Maximum number
of generations.
Fig.5 – Uniform crossover operator example.
This work used a probability pc of 0.7. An
example of this operator has been used successfully
in Gonçalves et al. [4].
5.3.4 Flat crossover
Consider the following parents:
Parent1 = (x1,1, …, x1,n)
Parent2 = (x2,1, …, x2,n)
and a vector of random values r = (r1, …, rn).
5.6 Evolutionary strategy
The offspring 1 = (x11, …, x1n) is computed as a
vector of linear combinations in the following way
(for all i = 1, …, n):
x1i  ri x1,i  (1  ri ) x2,i
,
E-ISSN: 2224-2872
i  1,..., n
To breed good solutions, the random key vector
population is operated upon by a genetic algorithm.
(5)
168
Issue 4, Volume 12, April 2013
WSEAS TRANSACTIONS on COMPUTERS
Jorge Magalhães-Mendes

There are many variations of genetic algorithms
obtained by altering the reproduction, crossover, and
mutation operators.
Reproduction is a process in which individual
(chromosome) is copied according to their fitness
values (makespan).
Reproduction is accomplished by first copying
some of the best individuals from one generation to
the next, in what is called an elitist strategy.
In this paper the fitness proportionate selection,
also known as roulette-wheel selection, is the genetic
operator for selecting potentially useful solutions for
reproduction. The characteristic of the roulette wheel
selection is stochastic sampling.
The fitness value is used to associate a
probability of selection with each individual
chromosome. If fi is the fitness of individual i in the
population, its probability of being selected is,
pi 
fi
,
N
f
i 1
i  1,..., n


FT6, FT 10 and FT20 originally proposed by
Fisher and Thompson [13];
LA1-LA40 proposed by Lawrence [14];
ORB1-ORB10, proposed by Applegate and
Cook [29].
Tables 1, 2, 3 and 4 summarize the experimental
results. It lists number of jobs, number of operations,
instance, best known solution (BKS), GA-MC-SPc,
GA-MC-TPc, GA-MC-Uc and GA-MC-Fc. The last
row of each table shows the value of the Average
Relative Deviation (ARD).
The ARD is calculated in the following way:
NIS
Cmax i  BKSi
i 1
BKSi
RE  
ARD 
(7)
RE
NIS
(8)
(6)
where NIS is number of instances solved.
i
A roulette wheel model is established to represent
the survival probabilities for all the individuals in the
population.
Fig. 7 shows this evolutionary strategy with the
operators selection, recombination (or crossover) and
mutation.
Table 1: Experimental results for
instances ABZ5-ABZ9.
Fig.7 – Evolutionary strategy.
Table 2: Experimental results for
instances FT06, FT10 and FT20.
This evolutionary strategy was applied to the
resource constrained project scheduling problem with
a good performance, see Mendes [2, 26].
In most of the problems, the single crossover
operator performs better than the others and is the
operator with lowest ARD, see Tables 1, 2, 3, 4 and
5.
Unlike the algorithm GA-MC-SPc that always
has the best ARD, the algorithm GA-MC-SPc has not
always the second best (lowest) ARD. Indeed, the
algorithm GA-MC-Fc for ABZ problems (Table 1)
obtains the second lowest ARD.
This result can lead to considerer that the
operator Fc is second best for larger problems, i.e.,
6 Numerical experiments
To evaluate the performance of each crossover
operator, we considered the following classes of
problems, taken from the literature:
 ABZ5, ABZ6, ABZ7, ABZ8 and ABZ9
proposed by Adams et al. [10];
E-ISSN: 2224-2872
169
Issue 4, Volume 12, April 2013
WSEAS TRANSACTIONS on COMPUTERS
Jorge Magalhães-Mendes
problems greater than 10x10. For problems larger
than 10x10, we obtain the following values of ARD:
1.51% (GA-MC-SPc), 1.89% (GA-MC-TPc), 1.84%
(GA-MC-Uc) and 2.13% (GA-MC-Fc), see Table 5.
Globally the second best value for the ARD is
obtained by algorithm GA-MC-Uc and the third best
value by the algorithm GA-MC-TPc, see Table 6.
With these results, for these types of problems,
the single-crossover performed the best results,
followed by the uniform crossover operator.
We must think why the SPc operator shows better
results. A strong possibility relates to the theory of
building blocks (Goldberg [1]). This operator has
only a cut, so it is more likely to not destroy blocks.
Instead, the operator of two points cut large blocks
and operator’s uniform and flat ignores its existence.
This can be a very important result in choosing
the crossover operator when using genetic algorithms,
particularly in job shop.
Table 4: Experimental results for
instances ORB1-ORB10.
Table 3: Experimental results for
instances LA01-LA40.
Table 5: Average relative deviation
for problems greater than 10x10.
Tables 1, 2, 3 and 4 shows that the GA-MC-SPc
is able to find the best know solution for 2 instances
in the set ABZ (40%), 3 instances in the set FT
(100%), 28 in the set LA (70%) and 9 in the set of
ORB (90%).
E-ISSN: 2224-2872
Additionally, the GA-MC-SPc results are
compared against the most recent work in the
literature.
The performance of GA-MC-SPc was evaluated
on a standard set of 50 benchmark instances
belonging to two classical sets known LA from
170
Issue 4, Volume 12, April 2013
WSEAS TRANSACTIONS on COMPUTERS
Jorge Magalhães-Mendes
Lawrence [14] and ORB from Applegate and Cook
[29].
authors have different approaches to solve the JSSP
problem.
We compare the obtained results with the
aforementioned state of art approaches:
 Qing-dao-er-ji and Wang [38];
 Rego and Duarte [8];
 Hasan et al. [27];
 Gonçalves et al. [4];
 Aarts et al. [15];
 Dorndorf et al. [18];
 Binato et al. [20];
 Nowicki and Smutnicki [31];
 Sha and Hsu [6];
 Pardalos et al. [32];
 Adams et al. [10];
 Zhang et al. [35].
The result is showed in Table 7.
The algorithm GA-MC-SPc gives good results
when compared to the state of art. In Table 7 we can
see that the algorithm proposed is among the top ten
best performing.
This study limited the number of generations for
all operators tested. The number of generations was
limited to 400. A higher number of generations for all
operators would increase the computational time.
Although there is convergence of the algorithms close
to 400 generations, an increase of generations in the
proposed algorithm can improve its global
performance.
The experiments were performed on an Intel
Core 2 Duo CPU T7250 @2.00 GHz. The
computational times dispended are in the range [7,
3100] seconds.
Table 6: Average relative deviation
with all problems.
As different authors used different number of
problems, the comparison is based only on those
problems that authors considered. Note that the
E-ISSN: 2224-2872
Table 7: Comparison of the %
deviations for the different number of
problems authors considered.
171
Issue 4, Volume 12, April 2013
WSEAS TRANSACTIONS on COMPUTERS
Jorge Magalhães-Mendes
Computers & Industrial Engineering, 51, 4, 2006, pp.
791-808.
[7] F. Pezzela and E. Merelli, A tabu search method
guided by shifting bottleneck for the job shop
scheduling problem, European Journal of Operational
Research, Vol. 120, 2000, pp. 297-310.
[8] C. Rego and R. Duarte, A filter-and-fan approach to
the job shop scheduling problem, European Journal of
Operational Research, Vol. 194, 2009, pp. 650–662.
[9] E. Nowicki and C. Smutnicki, A Fast Taboo Search
Algorithm for the Job-Shop Problem, Management
Science, Vol. 42, No. 6, 1996, pp. 797-813.
[10]J. Adams, E. Balas and Z. Zawack, The shifting
bottleneck procedure for job shop scheduling,
Management Science, Vol. 34, 1988, pp. 391-401.
[11] C. Y. Zhang, P. Li. and Z. Guan, A very fast TS/SA
algorithm for the job shop scheduling problem,
Computers & Operations Research, Vol. 35, 2008, pp.
282-294.
[12]H.R. Lourenço, Local optimization and the job-shop
scheduling problem, European Journal of Operational
Research, Vol. 83, 1995, pp. 347-364.
[13]H. Fisher and G.L. Thompson, Probabilistic Learning
Combinations of Local Job-Shop Scheduling Rules, in:
Industrial Scheduling, J.F. Muth and G.L. Thompson
(eds.), Prentice-Hall, Englewood Cliffs, NJ, 1963, pp.
225-251.
[14]S. Lawrence, Resource Constrained Project
Scheduling: An Experimental Investigation of Heuristic
Scheduling Techniques, GSIA, Carnegie Mellon
University, Pittsburgh, PA, 1984.
[15]E.H.L.Aarts, P.J.M.Van Laarhoven, J.K. Lenstra and
N.L.J.Ulder, A computational study of local search
algorithms for job shop scheduling, ORSA Journal on
Computing, 6, 1994, pp. 118-125.
[16]R.M.Aiex, S.Binato and M.G.C. Resende, Parallel
GRASP with Path-Relinking for Job Shop Scheduling,
Parallel Computing, Vol. 29, Issue 4, 2003, pp. 393 430.
[17]F. Croce, R. Tadei, and G. Volta, A Genetic
Algorithm for the Job Shop Problem, Computers and
Operations Research, Vol. 22(1), 1995, pp. 15-24.
[18]U. Dorndorf, and E. Pesch, Evolution Based Learning
in a Job Shop Environment, Computers and
Operations Research, Vol. 22, 1995, pp. 25-40.
[19]L. Wang, and D. Zheng, An effective hybrid
optimisation strategy for job-shop scheduling
problems, Computers & Operations Research, Vol. 28,
2001, pp. 585-596.
[20]S. Binato, W.J.Hery, D.M. Loewenstern and
M.G.C.Resende, A GRASP for Job Shop Scheduling.
In: Essays and Surveys in Metaheuristics, Ribeiro,
Celso C., Hansen, Pierre (Eds.), Kluwer Academic
Publishers, 2002.
[21]B. Giffler and G.L. Thompson, Algorithms for
Solving Production Scheduling Problems, Operations
Research, Vol. 8(4), 1960, pp. 487-503.
[22]J. Carlier and E. Pinson, An Algorithm for Solving the
Job Shop Problem. Management Science, Feb, 35(29),
1989, pp.164-176.
7 Conclusions and remarks
This paper presents an experimental study of a set of
crossover operators, namely single-crossover, twopoint crossover, uniform crossover and flat crossover.
The single-crossover has the best results, followed by
the uniform crossover operator.
This can be a very important result in choosing
the crossover operator when using genetic algorithms,
particularly in job shop.
Additionally, the best crossover operator was
tested on a set of 50 standard instances taken from the
literature and compared with the best state-of-the-art
approaches. The algorithm produced good results
when compared with other approaches.
Although there is convergence of the algorithms
close to 400 generations, an increase of generations in
the proposed algorithm (GA-MC-SPc) can improve
its global performance.
Further work could be conducted to explore the
possibility of genetically correct the chromosomes
supplied by the genetic algorithm to reflect the
solutions obtained by the local search heuristic.
Acknowledgements
This work has been partially supported by the
CIDEM - Centre for Research & Development in
Mechanical Engineering. CIDEM is a unit of FCT –
Portuguese Foundation for the Science and
Technology.
References:
[1]D. E. Goldberg. Genetic Algorithms in Search,
Optimization & Machine Learning, Addison-Wesley,
1989.
[2] J. Magalhães-Mendes, Project scheduling under
multiple resources constraints using a genetic
algorithm, WSEAS TRANSACTIONS on BUSINESS
and ECONOMICS, Issue 11, Volume 5, November
2008, pp. 487-496.
[3] I. Essafi, Y. Mati and S.D. Pérès, A genetic local
search algorithm for minimizing total weighted
tardiness in the job-shop scheduling problem,
Computers
&
Operations
Research,
Vol. 35, Issue 8, 2008, pp. 2599-2616.
[4] J.F. Gonçalves, J.M. Mendes, and M.C.G. Resende. A
hybrid genetic algorithm for the job shop scheduling
problem. European Journal of Operational Research,
Vol. 167, 2005, pp. 77-95.
[5] J.J.M. Mendes, J.F. Gonçalves and M.G.C. Resende,
A random key based genetic algorithm for the resource
constrained project scheduling problem. Comput.
Oper. Res. 36, 1, 2009, pp. 92-109.
[6] D. Y. Sha and C. Hsu, A hybrid particle swarm
optimization for job shop scheduling problem.
E-ISSN: 2224-2872
172
Issue 4, Volume 12, April 2013
WSEAS TRANSACTIONS on COMPUTERS
Jorge Magalhães-Mendes
[23]J. Carlier and E. Pinson, A practical use of Jackson’s
preemptive schedule for solving the job-shop problem.
Annals of Operations Research, Vol. 26, 1990, pp.
269-287.
[24]P. Brucker, B. Jurisch and B. Sievers, A Branch and
Bound Algorithm for Job-Shop Scheduling Problem,
Discrete Applied Mathematics, Vol. 49, 1994, pp. 105127.
[25]D. P.Williamson, L.A. Hall, J.A. Hoogeveen, , C. A.
J.Hurkens, J. K. Lenstra, S. V. Sevastjanov and D. B.
Shmoys, Short Shop Schedules, Operations Research,
45(2), 1997, pp. 288-294.
[26]J. Magalhães-Mendes, Project scheduling using a
competitive genetic algorithm. In Proceedings of the
8th WSEAS Conference on Simulation, Modelling and
Optimization (Santander, Cantabria, Spain). J. M. de la
Maza and P. L. Espí, Eds. Mathematics And
Computers In Science And Engineering. WSEAS,
Stevens Point, Wisconsin, 2008, pp. 39-42.
[27]S.M.K. Hasan, R. Sarker and D. Cornforth, GA with
Priority Rules for Solving Job-Shop Scheduling
Problems. In Proceedings of the IEEE Congress on
Evolutionary Computation CEC(2008), 2008, pp.
1913-1920.
[28]M. J. Varnamkhasti, L. S. Lee, M. R. Bakar, and W. J.
Leong, A Genetic Algorithm with Fuzzy Crossover
Operator and Probability, Advances in Operations
Research, vol. 2012, Article ID 956498, 16 pages,
2012. doi:10.1155/2012/956498
[29]D. Applegate and W. Cook, A computational study of
the job-shop scheduling problem. ORSA Journal on
Computing, 3(2), 1991, pp. 149-156.
[30]J. Magalhães-Mendes, A genetic algorithm for the job
shop scheduling with a new local search using Monte
Carlo method, In Proceedings of the 10th WSEAS
international conference on Artificial intelligence,
knowledge engineering and data bases (AIKED'11),
Cambridge, UK, 2011, pp. 26-31.
[31]E. Nowicki and C. Smutnicki, An Advanced Tabu
Search Algorithm for the Job Shop Problem, Journal
of Scheduling, 8, 2005, pp.145–159.
[32]P. Pardalos, O. Shylo and A. Vazacopoulos, Solving
job shop scheduling problems utilizing the properties
of backbone and big valley, Computational
Optimization and Applications, 47(1), 2010, pp.1-16.
[33]G. Sywerda, Uniform crossover in genetic algorithms,
in Proceedings of the 3rd International Conference on
Genetic Algorithms, 1989, pp. 2-9.
[34]K. A. De Jong and W. M. Spears, A formal analysis of
the role of multi-point crossover in genetic algorithms,
Annals of Mathematics and Artificial Intelligence, vol.
5, no. 1, 1992, pp. 1–26.
[35]C.Y. Zhang, P.G. Li, Z.L. Guan, and Y.Q. Rao. A tabu
search algorithm with a new neighborhood structure
for the job shop scheduling problem. Computers &
Operations Research, 34(11), 2007, pp. 3229-3242.
[36]Y.-Z. Luo, G.-J. Tang, Z.G. Wang and H.Y. Li.
Optimization of perturbed and constrained fuel-optimal
impulsive rendezvous using a hybrid approach.
Engineering Optimization, 38(8), 2006, pp. 959-973.
E-ISSN: 2224-2872
[37]R. R. Sharapov. Genetic Algorithms: Basic Ideas,
Variants and Analysis, Vision Systems: Segmentation
and Pattern Recognition, Goro Obinata and Ashish
Dutta (Ed.), 2007, pp. 407-422, ISBN: 978-3-90261305-9.
[38]R. Qing-doa-er-ji and Y. Wang. A new hybrid genetic
algorithm for job shop scheduling problem. Computers
& Operations Research, 39(10), 2012, pp. 2291-2299.
[39]H.R. Choi and B.J. Park, Genetic Algorithm for the
Integration of Process Planning and Scheduling in a
Job
Shop,
WSEAS
TRANSACTIONS
on
INFORMATION SCIENCE & APPLICATIONS, Issue
12, Volume 3, December 2006, pp. 2498-2504.
[40]H. Chiu, K. Hsieh, Y.T. Tang and W. Chien,
Employing a Genetic Algorithm Based on Knowledge
to Address the Job Shop Scheduling Problem, WSEAS
TRANSACTIONS on COMPUTER RESEARCH, Issue
2, Volume 2, February 2007, pp. 327-333.
[41]H. Chiu, K. Hsieh, Y.T. Tang and C.Y. Wang, A
Novel Approach to Address the Job-Shop Scheduling
Problem by using a Tabu Genetic Algorithm, WSEAS
TRANSACTIONS on COMPUTER RESEARCH, Issue
2, Volume 2, February 2007, pp. 339-345.
173
Issue 4, Volume 12, April 2013